Biology subjects

Bellazzi, R.

Publications and source records attributed to Bellazzi, R..

7 recordsLinked to original sources

bioRxiv2025

Vir2vec: A Genome-Wide Viral Embedding

Genomic language models (gLMs) are powerful numerical surrogates for DNA, but existing architectures focus primarily on human DNA or limited viral sets, and no dedicated benchmark exists for viral genome understanding. Here we introduce Vir2vec (from 17M to 422M parameters), decoder-only gLMs continually pretrained on a curated pan-viral corpus of 565,747 complete genomes across 295 species. We also present the viral Genome Understanding Evaluation (vGUE), a benchmark probing viral representations across broad organism discrimination, evolutionary signatures, intra-genus separation, variant subtyping, and host tropism. Evaluated as frozen feature extractors under nested cross-validation, Vir2vec achieves top balanced accuracy across 5 of 7 vGUE tasks, outperforming human-trained (MistralDNA), viral-specific (ModernBERTvirus), and broad biological foundation (Evo1) model. Furthermore, Vir2vec displays robust zero-shot generalization to unseen viral families. Finally, we introduce an argmax-routed SHAP explainability framework that projects feature attributions directly onto single-codon coordinates, correctly isolating key receptor-binding domain mutations driving mouse host adaptation. Vir2vec and vGUE provide a foundation for viral surveillance and discovery pipelines.

bioinformatics↗

bioRxiv2025

NanoCT-confocal mapping of mouse ovaries reveals lifespan-persistent symmetric organizational rules of primary-to-preovulatory follicles

Mammalian ovaries are organized into follicles integrated with the surrounding tissue, collectively shaping an architecture continuously remodeled through cycles of growth and elimination. Combining nano-Computed Tomography with 3D digital modeling, we deciphered spatial rules governing follicle organization in the mouse ovary across lifespan. We find that primary to preovulatory follicles adopt a quantitatively symmetric arrangement along the anterior-posterior and dorsal-ventral axes, established during prepuberty before secondary follicle vascularization and preserved through adulthood and aging, regardless of follicle health or atresia. A novel nanoCT-confocal pipeline revealed that follicles enclosing oocytes with transcriptionally-active or transcriptionally-inactive chromatin states follow divergent fates of growth, atresia, or developmental competence, yet retaining a symmetric distribution. Mathematical modeling of follicle growth and elimination from prepuberty to adulthood identified transition rates between follicle stages that support gonadal symmetry. These findings suggest an evolutionarily optimized developmental program ensuring follicle recruitment and high ovulatory output, a hallmark of polytocous species.

developmental biology↗

bioRxiv2024

SARITA: A Large Language Model for Generating the S1 Subunit of the SARS-CoV-2 Spike Protein

The COVID-19 pandemic has profoundly impacted global health, economics, and daily life, with over 776 million cases and 7 million deaths from December 2019 to November 2024. Since the original SARS-CoV-2 Wuhan strain emerged, the virus has evolved into variants such as Alpha, Beta, Gamma, Delta, and Omicron, all characterized by mutations in the Spike glycoprotein, critical for viral entry into human cells via its S1 and S2 subunits. The S1 subunit, binding to the ACE2 receptor and mutating frequently, affects infectivity and immune evasion; the more conserved S2, on the other hand, facilitates membrane fusion. Predicting future mutations is crucial for developing vaccines and treatments adaptable to emerging strains, enhancing preparedness and intervention design. Generative Large Language Models (LLMs) are becoming increasingly common in the field of genomics, given their ability to generate realistic synthetic biological sequences, including applications in protein design and engineering. Here we present SARITA, an LLM with up to 1.2 billion parameters, based on GPT-3 architecture, designed to generate high-quality synthetic SARS-CoV-2 Spike S1 sequences. SARITA is trained via continuous learning on the pre-existing protein model RITA. When trained on Alpha, Beta, and Gamma variants (data up to February 2021 included), SARITA correctly predicts the evolution of future S1 mutations, including characterized mutations of Delta, Omicron and Iota variants. Furthermore, we show how SARITA outperforms alternative approaches, including other LLMs, in terms of sequence quality, realism, and similarity with real-world S1 sequences. These results indicate the potential of SARITA to predict future SARS-CoV-2 S1 evolution, potentially aiding in the development of adaptable vaccines and treatments.

bioinformatics↗

bioRxiv2023

Forecasting dominance of SARS-CoV-2 lineages by anomaly detection using deep AutoEncoders

The coronavirus disease of 2019 (COVID-19) pandemic is characterized by sequential emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants, lineages, and sublineages, outcompeting previously circulating ones because of, among other factors, increased transmissibility and immune escape. We propose DeepAutoCoV, an unsupervised deep learning anomaly detection system to predict future dominant lineages (FDLs). We define FDLs as viral (sub)lineages that will constitute more than 10% of all the viral sequences added to the GISAID database on a given week. DeepAutoCoV is trained and validated by assembling global and country-specific data sets from over 16 million Spike protein sequences sampled over a period of about 4 years. DeepAutoCoV successfully flags FDLs at very low frequencies (0.01% - 3%), with median lead times of 4-17 weeks, and predicts FDLs [~]5 and [~]25 times better than a baseline approach For example, the B.1.617.2 vaccine reference strain was flagged as FDL when its frequency was only 0.01%, more than a year before it was considered for an updated COVID-19 vaccine. Furthermore, DeepAutoCoV outputs interpretable results by pinpointing specific mutations potentially linked to increased fitness, and may provide significant insights for the optimization of public health pre-emptive intervention strategies. Key PointsO_LIIntroduction of DeepAutoCoV: The article introduces DeepAutoCoV, an unsupervised deep learning anomaly detection system designed to predict future dominant lineages (FDLs) of SARS-CoV-2. FDLs are defined as viral (sub)lineages that will constitute more than 10% of all viral sequences added to the GISAID database in a given week; C_LIO_LIPerformance and Predictive Capability: DeepAutoCoV successfully flags FDLs at very low frequencies (0.01% to 3%), with median lead times of 4 to 17 weeks before they become dominant. It predicts FDLs approximately 5 to 25 times better than baseline approaches. For instance, the B.1.617.2 vaccine reference strain was identified when its frequency was only 0.01%, over a year before it was considered for vaccine updates; C_LIO_LIInterpretable Results and Mutation Identification: The system provides interpretable results by pinpointing specific mutations that may be linked to increased fitness, offering insights that can optimize public health interventions. Key FDL mutations, such as those found in Delta and Omicron variants, are identified and analyzed for their potential impact on viral spread and immune escape; C_LIO_LIAdvantages and Applications: DeepAutoCoV is advantageous because it does not require prior assumptions about which protein sites are more likely to mutate. Its application in genomic surveillance systems could significantly reduce the time needed for public health responses to emerging variants, enabling early interventions such as vaccine updates; C_LIO_LIEvaluation and Comparisons: The performance of DeepAutoCoV was tested over four years of global and national surveillance data, demonstrating superior predictive power compared to other supervised and unsupervised methods. The system is periodically updated to adapt to the evolving viral landscape, making it a robust tool for ongoing surveillance efforts. C_LI

bioinformatics↗

bioRxiv2023

Digenic variant interpretation with hypothesis-driven explainable AI

MotivationThe digenic inheritance hypothesis holds the potential to enhance diagnostic yield in rare diseases. Computational approaches capable of accurately interpreting and prioritizing digenic combinations based on the probands phenotypic profiles and familial information can provide valuable assistance to clinicians during the diagnostic process. ResultsWe have developed diVas, a hypothesis-driven machine learning approach that can effectively interpret genomic variants across different gene pairs. DiVas demonstrates strong performance both in classifying and prioritizing causative pairs, consistently placing them within the top positions across 11 real cases (achieving 73% sensitivity and a median ranking of 3). Additionally, diVas exploits Explainable Artificial Intelligence (XAI) to dissect the digenic disease mechanism for predicted positive pairs. Availability and ImplementationPrediction results of the diVas method on a high-confidence, comprehensive, manually curated dataset of known digenic combinations are available at oliver.engenome.com.

bioinformatics↗

bioRxiv2020

Harnessing cytoplasmic particles movement of the human early embryo analysed by advanced imaging and artificial intelligence to predict development to blastocyst stage

Research QuestionProgress in artificial intelligence (AI) and advanced image analysis offers unique opportunities to develop novel embryo assessment approaches. In this study, we tested the hypothesis that such technologies can extract and harness novel information derived from cytoplasmic movements of the early human embryo to predict development to blastocyst. DesignIn a proof-of principle study, an artificial neural network (ANN) approach was undertaken to assess retrospectively 230 human preimplantation embryos. After ICSI, embryos were subjected to time-lapse monitoring for 44 hours. For comparison as a standard embryo assessment methodology, a single senior embryologist assessed each embryo to predict development to blastocyst stage (BL) based on a single picture frame taken at 42 hours of development. In the experimental approach, in embryos that developed to blastocyst or destined to arrest (NoBL), cytoplasm movement velocity (CMV) was recorded by time-lapse monitoring during the first 44 hours of culture and analysed with a Particle Image Velocimetry (PIV) algorithm to extract quantitative information. Three main AI approaches, the k-Nearest Neighbor (k-NN), the Long-Short Term Memory Neural Network (LSTM-NN) and the hybrid ensemble classifier (HyEC) were employed to classify the two embryo classes. ResultsBlind operator assessment classified each embryo in terms of ability of development to blastocyst, reaching a 75.4% accuracy, 76.5% sensitivity, 74.3% specificity, 74.3% precision and 75.4% F1 score. After integration of results from AI models together with the blind operator classification, the performance metrics improved significantly, with a 82.6% accuracy, 79.4% sensitivity, 85.7% specificity, 84.4% precision and 81.8% F1 score. ConclusionsThe present study suggests the possibility to predict human blastocyst development at early cleavage stages by detection of CMV and AI analysis. This indicates the importance of the dynamics of the cytoplasm as a novel and valuable source of data to assess embryo viability.

developmental biology↗

bioRxiv2019

Autologous micrograft accelerates endogenous wound healing response via ERK-induced cell migration

Defective fibroblast migration causes delayed wound healing (WH) and chronic skin lesions. Autologous micrograft (AMG) therapies have recently emerged as a new effective treatment able to improve wound healing capacity. However, the molecular mechanisms connecting their beneficial outcomes with the wound healing process are still unrevealed. Here, we show that AMG modulates primary fibroblast migration and accelerates skin re-epithelialization without affecting cell proliferation. We demonstrate that AMG is enriched in a pool of WH-associated growth factors that may provide the initiation signal for a faster endogenous wound healing response. This, in turn leads to increased cell migration rate by elevating activity of extracellular signal-regulated kinase (ERK) pathway and subsequent activation of matrix metalloproteinase expression and their extracellular enzymatic activity. Moreover, AMG-treated wounds showed increased granulation tissue formation and organized collagen content. Overall, we shed light on AMG molecular mechanism supporting its potential to trigger a highly improved wound healing process.

molecular biology↗